60 
• Ions from different formula groups including their isotope signals have the same 
nominal mass and overlapping retention time ranges and can therefore cause 
interferences, when low resolution mass spectrometry is applied. 
First methods for PC As analysis were developed during the 1980ies (Gjps and 
Gustavsen, 1982; Schmid el al., 1985). Electron capture negative ion (ECNI) detection 
was selected due to strongly decreased detection limits caused by restricted molecule 
fragmentation. This technique is based on the formation of anions due to the high 
electron affinity of polychlorinated compound by capture of electrons (in thermal 
equilibrium). Matrix compounds and other interferences have a low electron affinity and 
do not form anions (Oehme, 1998). Moreover, high resolution mass spectrometry 
(HRMS) was chosen to suppress disturbances from PCAs themselves as well as from 
other not completely removed polychlorinated pollutants such as PCB (Tomy et al., 
1997). ECNI was still the method of choice when this project was started in 2002 
(Froescheis and Ballschmiter, 1998). 
• However, ECNI generates additional problems influencing the reliability of 
results (Reth and Oehme, 2004; Reth et al., 2005): 
• Response factors are strongly dependent on the number of Cl atoms and could 
vary by one order of magnitude. Moreover, co-elution of compounds with 
different degree of chlorination can influence the overall response. 
• Lower chlorinated isomers with <3-4 Cl atoms have low response factors and 
could normally not be detected. 
• Molecular ions M"' as well as [M-Cl]" and the chlorine adduct [M+C1] + are formed 
simultaneously and increased the risk of interfering mass overlap. 
• The overall degree of chlorination influences the overall response factor of the 
PCA mixture present in an environmental sample and required references with 
similar composition for quantification, which normally are not available. 
Nevertheless, the quantification procedures developed of Tomy et al. (1997; Tomy and 
Stern, 1999) allowed to minimise the risk of systematic errors and gave rather reliable 
results. However, this required HRMS as well as many runs up of the same sample (up 
to twenty) to detect all possible formula groups (chain lengths from Cio n and number